Do you know how many hurricanes there will be in the Atlantic next year? By how much global sea levels will rise over 20 years? What average temperature increase the Earth will experience this century? If you answered with an unqualified “Yes” or “No” to any of these questions, then I might suggest you are either a crank or a clairvoyant. The fact is, none of us know for certain. We can, however, make calculated estimates of likelihood – and the more expertise you have and the more careful your thinking and modelling, the better those probability estimates are likely to be. And there are multiple groups around the world who value this kind of information greatly. Insurance companies want to know how much money they might be expected to pay out on claims resulting from hurricane or flood damage. Governments want to be able to prepare and adapt their countries for what lies ahead. Farmers desire information on how their crop harvest might change. Add in supermarkets, consumers, people living in low-lying countries and wildfire-prone areas, along with many other interested parties, and you have a thirst for accurate probabilistic predictions. This is where we come in with the Climate Risk and Uncertainty Collective Intelligence Aggregation Laboratory (CRUCIAL). MAKE A PREDICTION I am an economist, but the applications of economics extend further than you might imagine. I want to see beyond stereotypical applications, and prediction markets such as those we provide through CRUCIAL do that. Prediction markets do just what they say – bring people together to predict the probability of what is going to happen in the future. Since we do not have certainty about the future, our expectations and beliefs are central in operationalising financial or policy decision-making questions, including those pertaining to sustainability and climate change. But when you tell people you are working on a prediction market, what immediately comes to mind is sports betting or political prediction markets. Those markets attract enough attention and a sufficiently passionate following of willing uninformed bettors to make it worthwhile for well-informed experts to participate when the market is organised in a ‘pay-to-play’ format. Aside from the ethical considerations of this format, the not-so-small wrinkle is that it requires a gambling licence! For these reasons, we do not want to use these conventional frameworks. The distinction between our prediction markets and other similar types of markets, is that ours are set up specifically to incentivise the elicitation and aggregation of probabilistic information. We are not open access. You must be invited, and we set certain expertise criteria. We have representatives from some of the best academic departments in the world. Rather than being pay-to-play, the markets are pre-funded by ourselves, by interested organisations, or, in the future, by a consortium of organisations. EXPERTS ASSEMBLE Rather than typical climate risk forecasting, which relies on one team – often the team that agreed to do the forecasting at the lowest price – we bring together teams using different modelling and forecasting methodologies. You could use machine learning, artificial intelligence, traditional climate modelling techniques, or econometrics, for instance. If you use a different method, a slightly different econometric or statistical technique, your predictive probability distribution – the range and distribution of likelihood– will be somewhat different. But we believe this diversity is important. It is not only the modelling frameworks. There are dozens upon dozens of decisions to be made to implement the analysis. How do we validate and process the data? What do we do with missing observations and outliers? So, there is no surprise that the resulting probability distributions are different. The question is how do you aggregate those forecasts? What relative weights do you place on those rigorous but different forecasts? 40 |
RkJQdWJsaXNoZXIy NTI5NzM=